angie-conv-image/scripts/overlay-dir-list.sh

#!/bin/sh
set -f

## ephemeral limit, keep in sync with verify_depth()
max_depth=99
default_depth=20

me="${0##*/}"
usage() {
	cat >&2 <<-EOF
	# usage: ${me} [options] <directory> [<directory> ..]
	# options:
	#  -d <N>, --depth <N>, --depth=<N>
	#    limit search depth to <N> (default: ${default_depth}, max: ${max_depth})
	#  -z, --zero
	#    separate entries with NUL instead of LF
	#  -s, --split-path
	#    separate directory and file name parts with "|" instead of "/"
	EOF
	exit "${1:-0}"
}
[ $# != 0 ] || usage

msg() { echo ${1:+"# ${me}: $*"} >&2 ; }
msgf() {
	_____fmt="$1" ; shift
	env printf "# ${me}: ${_____fmt}\\n" ${@+"$@"} >&2
	unset _____fmt
}

verify_depth() {
	case "$1" in
	[1-9] | [1-9][0-9] ) return 0 ;;
	esac
	msgf 'error: wrong depth specifier: %q' "$1"
	usage 1
}

o_depth=
f_zero_eol=
f_split_path=

## process options
want_value=
n_opt=0
for i ; do
	if [ -n "${want_value}" ] ; then
		case "${want_value}" in
		depth )
			o_depth="$i"
			verify_depth "${o_depth}"
		;;
		esac

		want_value=
		n_opt=$((n_opt+1))
		continue
	fi

	case "$i" in
	-d | --depth | --depth=* )
		if [ -n "${o_depth}" ] ; then
			msg 'error: "depth" option already set'
			usage 1
		fi
		case "$i" in
		*=* )
			o_depth="${i#*=}"
			verify_depth "${o_depth}"
		;;
		* ) want_value=depth ;;
		esac
	;;
	-z | --zero )
		if [ -n "${f_zero_eol}" ] ; then
			msg 'error: "zero" flag already set'
			usage 1
		fi
		f_zero_eol=1
	;;
	-s | --split-path )
		if [ -n "${f_split_path}" ] ; then
			msg 'error: "split-path" flag already set'
			usage 1
		fi
		f_split_path=1
	;;
	-* )
		msgf 'unknown option: %q' "$i"
		usage 1
	;;
	* ) break ;;
	esac

	n_opt=$((n_opt+1))
done

[ ${n_opt} = 0 ] || shift ${n_opt}

[ $# != 0 ] || usage 1

: "${o_depth:=${default_depth}}"

path_sep='/'
[ -z "${f_split_path}" ] || path_sep='|'

entry_sep='\n'
[ -z "${f_zero_eol}" ] || entry_sep='\0'

entry_fmt="%s${path_sep}%s${entry_sep}"

## work directory
w=$(mktemp -d) ; : "${w:?}"

_cleanup() {
	cd /
	rm -rf -- "$w"
}

adjust_with_skiplist() {
	[ -s "$1" ] || return 0
	[ -s "$2" ] || return 0

	__skiptype="$3"
	while read -r __skipname ; do
		[ -n "${__skipname}" ] || continue

		mawk \
		  -v "n=${__skipname}" \
		  -v "t=${__skiptype}" \
		'
			BEGIN {
				RS = ORS = "\0";
				FS = OFS = "|";
			}
			{
				if ($3 != n) { print; next; }
				if (t == "") { next; }
				if ($1 ~ t) { next; }
				print;
			}
		' < "$2" > "$2.t"
		mv -f "$2.t" "$2"
	done < "$1"
	unset __skipname __skiptype
}

: > "$w/all"

## process arguments
for i ; do
	## arguments with '|' are skipped - try naming paths simplier
	case "$i" in
	*\|* )
		msgf 'argument with "|" is IGNORED: %q' "$i"
		continue
	;;
	esac

	## non-existent directories are silently skipped
	[ -d "$i" ] || continue

	## generate directory listing
	## paths with '\n' and '|' are silently skipped - try naming paths simplier
	find "$i/" -follow -mindepth 1 -maxdepth 1 ! -name '*|*' -printf '%y|%h|%P\0' \
	| sed -zE '/\n/d' > "$w/current"

	## filter out uncommon entry types
	## allowed types:
	##   d - directory
	##   f - file
	## less common types (but also allowed):
	##   b - block device
	##   c - character device
	##   p - pipe
	##   s - socket
	sed -i -zE '/^[^bcdfps]/d' "$w/current"

	## empty directory - continue with next argument
	[ -s "$w/current" ] || continue

	## generate skiplist
	grep -zE '\.-$' < "$w/current" \
	| cut -z -d '|' -f 3 \
	| sed -zE 's/\.-$//' \
	| tr '\0' '\n' > "$w/skip"
	## adjust current list: remove entries from skiplist
	sed -i -zE '/\.-$/d' "$w/current"

	## adjust accumulated list: remove "skip" entries
	adjust_with_skiplist "$w/skip" "$w/all"
	rm -f "$w/skip"

	## adjust accumulated list: override all entries with "non-dir" entries from current list
	## NB: entry type from current list overrides entry from previously accumulated list
	sed -zEn '/^[^d]/p' < "$w/current" \
	| cut -z -d '|' -f 3 \
	| tr '\0' '\n' > "$w/nondir"
	adjust_with_skiplist "$w/nondir" "$w/all"
	rm -f "$w/nondir"

	## adjust accumulated list: override "non-dir" entries with "dir" entries from current list
	## NB: entry type from current list overrides entry from previously accumulated list
	sed -zEn '/^d/p' < "$w/current" \
	| cut -z -d '|' -f 3 \
	| tr '\0' '\n' > "$w/dir"
	adjust_with_skiplist "$w/dir" "$w/all" '[^d]'
	rm -f "$w/dir"

	## merge lists
	sort -zV -t '|' -k 3 < "$w/current" >> "$w/all"
	rm -f "$w/current"
done

# nothing to do?
if ! [ -s "$w/all" ] ; then
	_cleanup
	exit 0
fi

cut -z -d '|' -f '3' < "$w/all" \
| sort -zuV \
| tr '\0' '\n' > "$w/names"

: > "$w/dirnames"

sub_depth=$(( o_depth - 1 ))

while read -r name ; do
	mawk \
	  -v "n=${name}" \
	'
		BEGIN {
			RS = "\0"; ORS = "\n";
			FS = OFS = "|";
		}

		$3 == n { print; }
	' < "$w/all" > "$w/list"

	while IFS='|' read -r dtype dir _name ; do
		[ -n "${dtype}" ] || continue

		# ${_name} is unused
		case "${dtype}" in
		d )
			[ "${sub_depth}" != 0 ] || continue

			if grep -Fxq -e "${name}" "$w/dirnames" ; then
				continue
			fi

			printf '%s\n' "${name}" >> "$w/dirnames"

			mawk \
			  -v "n=${name}" \
			'
			BEGIN {
				ORS = "\0";
				FS = "|"; OFS = "/";
			}

			$3 == n { print $2,$3; }
			' < "$w/list" > "$w/dir.args"

			xargs -0 -a "$w/dir.args" "$0" ${f_zero_eol:+-z} ${f_split_path:+-s} -d "${sub_depth}"
			rm -f "$w/dir.args"
		;;
		* )
			printf "${entry_fmt}" "${dir}" "${name}"
		;;
		esac
	done < "$w/list"

done < "$w/names"

_cleanup
exit 0